Many oral squamous cell carcinomas are preceded by a precancerous stage characterized histologically by varying degrees of epithelial dysplasia. However, in a large proportion of cases, patients with oral epithelial dysplasia do not progress to oral cancer and prediction which lesions will progress is not possible by conventional clinical or histological methods. The objective of the present case control study is to evaluate whether changes in patterns of inactivation of the CDKN2A gene present in oral epithelial dysplasia may predict progression to oral cancer. The CDKN2A gene locus encodes two different transcripts derived from alternative splicing. P16 INK4A (exons alpha,2,3) acts as a critical G1 cell cycle inhibitor via CDK-4 and -6. P14/ARF (exons 1beta,2,3) acts to modulate MDM2 mediate degradation of the tumour suppressor gene p53. From the Oral Cancer Histopathology Core we have identified two groups of 2o patients with precancerous oral lesions. One group has progressed to oral cancer and the other has remained in the precancerous stage after many years of follow-up. Sequential biopsies have been obtained from the same intra-oral site in each patient and clinical in formation including smoking status recorded. This is exceptionally difficult material to collect and thus provides a unique opportunity to examine multistep carcinogenesis in vivo. We will use molecular studies and then compare the timing and patterns of CDKN2A (p16/INK4A and p14/ARF) gene inactivation and protein expression in sequential biopsies of epithelial dysplasias that have progressed to oral cancer with those that have not progressed to cancer. Results from these studies will determine changes in this critical tumour suppressor gene necessary for the transformation of oral precancerous lesions to cancer. Moreover, they offer the possibility of molecular assays to predict progression to oral cancer for the individual patient.